| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Department of Anesthesiology, Hospital for Special Surgery and the Weill College of Medicine of Cornell University, New York, New York; and Department of Anesthesiology and Critical Care Medicine, The Johns Hopkins University, Baltimore, Maryland.
Address correspondence to Spencer S. Liu, MD, Department of Anesthesiology, Hospital for Special Surgery, 535 East 70th St., New York, NY 10021. Address e-mail to liusp{at}hss.edu.
Abstract
BACKGROUND: Few individual clinical trials have had sufficient subject numbers to definitively determine the effects of postoperative analgesia on major outcomes.
METHODS: We systematically searched the Medline and the Cochrane Library databases for the past decade and focused on meta-analyses and large, randomized, controlled trials.
RESULTS: Eighteen meta-analyses, 10 systematic reviews, 8 additional randomized, controlled trials, and 2 observational database articles were identified for review or comment. Epidural analgesia with local anesthetics has the greatest theoretical potential to affect major outcomes and has been the most thoroughly investigated technique. The majority of evidence favors an ability of epidural analgesia to reduce postoperative cardiovascular and pulmonary complications only after major vascular surgery or in high-risk patients. This finding may become irrelevant because of rapid conversion of major surgery to minimally invasive techniques (e.g., endoluminal abdominal aortic repair) that carry less risk of complications. There is also consistent evidence that epidural analgesia with local anesthetics is associated with faster resolution of postoperative ileus after major abdominal surgery. Again, this finding may also become irrelevant with the adoption of laparoscopic techniques and multimodal fast-track programs for abdominal surgery. There is no current evidence that perineural analgesia, continuous wound catheters using local anesthetics, IV patient-controlled analgesia with opioids, or addition of multimodal systemic analgesics have any clinically significant beneficial effect on postoperative complications.
CONCLUSIONS: Overall, there is insufficient evidence to confirm or deny the ability of postoperative analgesic techniques to affect major postoperative mortality or morbidity. This is primarily due to typically insufficient subject numbers to detect differences in currently low incidences of postoperative complications.
Provision of high-quality postoperative analgesia has become recognized as an important perioperative goal. The potential to modify major postoperative complications by providing high-quality postoperative pain control has been a popular area for investigation in the past decade with more than 3800 clinical trials indexed in the National Library of Medicine’s Medline database (through June 2006). Despite the popularity of the topic, consensus on effects of acute postoperative pain control on outcomes remains controversial. A key component of this continuing controversy is the need for large patient numbers in any individual clinical trial due to the current relatively low incidences of major postoperative morbidity. For example, an observational meta-analysis in 2003 (1) of 176 studies enrolling more than 205,000 subjects undergoing the traditionally "high-risk" procedure of coronary artery bypass graft (CABG) surgery reported a mortality rate of approximately 1.7%. The same has become true for high-risk populations, such as the elderly. A random 5% sample of Medicare insurance claims (1997–2001) collected postoperative data from 68,726 patients undergoing a variety of surgical procedures and noted a 30-day mortality rate of 2.5% (2). Such modest incidences require approximately 4600 patients in a single, randomized, controlled trial (RCT) to be able to detect a 50% reduction in incidence from 2% to 1%. Even the highest reported rates of postoperative morbidity, such as pulmonary complications after thoracic or abdominal surgery (23%–36% incidence), would require approximately 100 patients per group to detect a 50% reduction (3–5). There are very few individual RCTs to definitively answer such questions. Thus, we have systematically reviewed the literature, and focused on studies with large subject samples, such as meta-analyses, to acquire suitable evidence on the effects of acute postoperative analgesia on major postoperative outcomes.
METHODS FOR SYSTEMATIC REVIEW
The National Library of Medicine’s Medline and the Cochrane Library databases were searched for the past decade (January 1996–June 1, 2006). The search was limited to the last decade because of evolving surgical techniques, trends in perioperative care, and trends in postoperative analgesia. Both authors performed independent searches. The following search terms were used in various combinations: Postoperative pain, Postoperative analgesia, Epidural analgesia, Nerve block, Catheters, Patient controlled analgesia, NSAIDs, Acetaminophen, Ketamine, Multimodal, Postoperative complications, Mortality, Myocardial infarction, Myocardial ischemia, Heart failure, Dysrhythmias, Pneumonia, Pulmonary complications, Respiratory insufficiency, Ileus, Infection, Wound infection, Surgical wound infection, Delirium, Cognitive disorders, Cognition, Cognition disorders, Deep venous thombosis, Pulmonary embolus, Vascular surgery, Blood vessel prosthesis, Chronic pain, Intractable pain, Epidemiology, and Databases. Searches were limited to "Human," "Clinical trials," "Randomized controlled trial," "Review article," "Meta-analysis" or "Evidence based review." Additional articles were also sought from the authors’ personal files.
Inclusion criteria and data extraction: All the above abstracts were reviewed for potential inclusion in the systematic review. To acquire large patient numbers, preference was given to the most recent meta-analyses and only articles published since those quantitative reviews were considered for individual inclusion. If more than one meta-analysis was identified, then preference was given to the most recent to avoid overlap. Because examination of studies with large sample sizes was the goal of this review, we selected an n 
200 (100 per randomized group) as our inclusion criteria for individual RCTs based on an ability to detect a 50% reduction in incidence from 34% to 17%. This inclusion criterion was based on highest reported incidences of postoperative complications to maximize capture of RCTs. Previous studies have reported incidences of postoperative pulmonary complications (PPC) as high as 23%–36%, depending on definition after thoracic and abdominal surgery in high risk patients (3–5). Thus, we selected the high end of the range and then rounded to an even number of subjects. These inclusion criteria identified 18 meta-analyses, 10 systematic reviews, 8 additional RCTs, and 2 observational database articles for review or comment. Data are displayed in tables for qualitative review.
EFFECT OF POSTOPERATIVE ANALGESIC TECHNIQUE ON MAJOR OUTCOMES
Regional Analgesia
Epidural Analgesia
Mortality
Meta-Analyses.
The largest meta-analysis of RCTs (6) was published in 2000 (CORTRA). This meta-analysis included 141 RCTs (through January 1, 1997) with 9559 patients undergoing a variety of surgical procedures (Table 1). Although the intent of this meta-analysis was to examine the effects of neuraxial block (spinal anesthesia, epidural anesthesia, and epidural analgesia) versus general anesthesia, results from this meta-analysis likely apply to the epidural analgesia subset as 66 of the RCTs with 4498 of the patients using epidural anesthesia and analgesia. This meta-analysis observed a reduction in mortality with neuraxial blockade (1.9% vs 2.8%, OR 0.7 with 95% CI 0.54–0.9) and specifically for thoracic epidural blocks (1.5% vs 2.9%). This reduction was attributed to reduction of adverse outcomes in multiple organ systems that will be discussed individually. No specific surgical procedures were identified as benefiting from central neuraxial block other than the general class of orthopedics. An earlier meta-analysis (2001) by Beattie et al. (8) more specifically examined RCTs for use of postoperative epidural analgesia for 24 h or more after surgery. This meta-analysis identified 11 RCTs (through 1998) with 1173 patients undergoing mixed but primarily major vascular surgery. The incidence of mortality was nonsignificantly reduced with epidural analgesia (3.1% vs 4.4%, P = 0.3), but this study lacked adequate sample size.
|
|
Randomized Controlled Trial.
Since the publication of the CORTRA meta-analysis in 2000, two large multicenter RCTs examining epidural analgesia have been published (Table 1). In 2001, the Veterans Affairs Cooperative Studies Program (VACS) randomized 984 patients (all or mostly men) undergoing four types of surgery (aortic, gastric, biliary, or colon) to combined general/epidural anesthesia followed by epidural morphine for an unspecified time versus general anesthesia followed by systemic opioid treatment (12). Approximately 85% of the epidurals were placed at the thoracic level. Overall mortality rates were similar between groups (4% vs 3.4%), but the study was not adequately sized to assess mortality. In 2002, the Multicenter Australian Study of Epidural Anesthesia (MASTER) trial was published (13). This study enrolled 915 high-risk patients (prospectively defined in the protocol) who had undergone mixed abdominal surgical procedures and were randomized to combined general/epidural anesthesia followed by 72 h of postoperative epidural analgesia (low thoracic or high lumbar placement) with local anesthetic and opioids versus general anesthesia followed by systemic opioid treatment (IV patient-controlled analgesia (PCA) or prn). This RCT was limited by poor protocol compliance, as only 225/447 patients fully adhered to the epidural analgesia protocol. Overall mortality rates were again similar between groups (5.1% vs 4.3%), but the study was also not adequately sized to assess mortality.
Medicare Database.
Large-scale observational databases offer the ability to acquire data on large numbers of patients (Table 1). In 2004, a 5% random sample of the Medicare claims database from 1997 through 2001 was analyzed (2). Patients undergoing a variety of surgical procedures (colectomy, esophagectomy, gastrectomy, hysterectomy, liver resection, nephrectomy, pulmonary resection, radical retropubic prostatectomy, total knee replacement, Whipple) were stratified according to the presence (n = 12,780 subjects) or absence (n = 55,943) of a bill for postoperative epidural analgesia. After adjusting for comorbidities, age, gender, and hospital size, regression analysis revealed that the presence of postoperative epidural analgesia was associated with a significantly lower incidence for both 7-day (0.5% vs 0.8%, OR = 0.52 with 95% CI 0.38–0.73) and 30-day (2.1% vs 2.5%, OR = 0.74 with 95% CI 0.63–0.89) mortality (2). Not unexpectedly, there was a significantly lower mortality in patients who received postoperative epidural analgesia for higher-risk procedures (e.g., lung resection, colectomy) but no difference in mortality between patients who did or did not receive postoperative epidural analgesia in lower-risk procedures (e.g., total knee replacement, hysterectomy) and when patients also had lower comorbidity indices. No mechanism for reduction in mortality with epidural analgesia was identified, as no reductions in other types of morbidity were observed (See sections below). A separate analysis of Medicare claims only in patients undergoing total hip replacement between 1994 and 1999 with (n = 2591) and without (n = 20,545) epidural analgesia also noted a nonsignificant reduction in mortality (0.2% vs 0.4%, OR 0.6 with 95% CI 0.2–1.5) in this relatively low-risk procedure (14). Although the number of patients from these database analyses is impressive, these data particularly suffer from methodological issues such as the retrospective nature, accuracy of coding for complications, and degree of association between epidural analgesia and outcomes.
Summary Statement.
There is poor evidence for reduction of mortality with epidural analgesia (Table 1). The largest meta-analysis observed a reduction with neuraxial block but examined a combination of spinal anesthesia, epidural anesthesia, and epidural analgesia. Thus, findings may not apply to the subset of epidural analgesia. Procedure-specific meta-analyses and individual RCTs have not noted an effect from epidural analgesia but lack sufficient sample size because of the relatively low incidence of mortality (0.2%–5%). Analysis of the Medicare claims database offers large patient numbers and an association between epidural analgesia and reduced mortality, but the methodology is limited.
Cardiovascular.
Approximately 100 million adults worldwide undergo noncardiac surgery annually, and nearly half of the patients are estimated to have cardiac risk factors (15). As such, it is estimated that nearly 500,000–900,000 patients will suffer a perioperative cardiovascular complication (15). Reported incidences of perioperative cardiovascular complications vary depending on surgical procedure and patient population. The highest incidences are typically reported for emergency surgery in the elderly, which may carry a very high risk when compared with elective surgery (19% vs 0.2% for myocardial infarction) (2,16). High-risk elective procedures are typically considered to be cardiac surgery (<2% incidence of myocardial infarction) and major vascular surgery (5%–10% incidence) (1,17). The economic burden of perioperative cardiovascular complications is also high, as more patients require admission to intensive care units, hospital stay is typically prolonged by 11 days and cost is increased by nearly 10,000 USD per patient (15,17). In the United States, this would approximate an increased cost of 444 million USD from major vascular procedures alone (17).
Uncontrolled postoperative pain may contribute to cardiac morbidity through activation of the sympathetic nervous system, surgical stress response, and coagulation cascade. Increased sympathetic nervous system activity can increase myocardial oxygen demand by increasing heart rate, arterial blood pressure, and contractility. In addition, sympathetic activation may enhance perioperative hypercoagulability, which may contribute to perioperative coronary thrombosis or vasospasm, thus reducing myocardial oxygen supply (18,19). Use of perioperative ß-adrenergic blockade to reduce the effects of this sympathetic activation has been demonstrated to improve clinical outcomes and is a popular strategy in high-risk patients (20).
Experimental data suggest that thoracic epidural anesthesia with local anesthetics can reduce sympathetic activation and provide a favorable balance of myocardial oxygen. In humans, thoracic epidural anesthesia may increase myocardial oxygen supply by selectively increasing the diameter of stenotic epicardial coronary arteries in patients with coronary artery disease while maintaining coronary perfusion pressure (18). The ability of thoracic epidural anesthesia to increase myocardial blood flow in patients with multivessel ischemic heart disease has been demonstrated even in the presence of sympathetic stimulation (21). In addition to improving myocardial oxygen supply, thoracic epidural anesthesia may decrease myocardial oxygen demand by decreasing pain, heart rate, and systemic vascular resistance (18). The combination of reduced sympathetic activity, improved oxygen supply, and decreased demand may explain the successful ability of thoracic epidural anesthesia to treat medically refractory angina (22). All these studies primarily used epidural anesthesia with larger doses of local anesthetics than the dilute solutions of local anesthetic and opioid typically used for postoperative thoracic epidural analgesia (TEA). Lumbar epidural anesthesia may not provide the same physiologic benefits as thoracic epidural anesthesia. Experimental studies have reported a compensatory increase in sympathetic activity above the level of blockade for lumbar epidural analgesia (LEA) (23), and clinical studies have noted increased incidences of left ventricular wall dysfunction with lumbar versus thoracic epidural anesthesia (18). However, the analgesic effects of LEA may still play a protective role, as a small study (n = 68) noted marked reduction in cardiovascular events (0% vs 19%) in patients with hip fractures randomized to preoperative LEA versus systemic analgesia (16).
Meta-Analyses.
Four meta-analyses were identified that examined the efficacy of epidural analgesia on cardiovascular events (Table 1) (1,6,8). The largest was the previously described CORTRA meta-analysis that examined neuraxial block (spinal anesthesia, epidural anesthesia, and epidural analgesia) versus general anesthesia. This meta-analysis reported a nonsignificant decrease in the risk of myocardial infarction (0.9% vs 1.3%). It should be noted that the majority of patients received lumbar epidural or spinal anesthesia, which, as previously mentioned, may not provide the physiologic benefit of TEA.
Three smaller but more specific meta-analyses examining the efficacy of postoperative epidural analgesia and cardiovascular events suggest a benefit for epidural analgesia and TEA in particular. Beattie et al. (8) identified RCTs (through December 1998) for mixed surgical procedures in which epidural analgesia was extended at least 24 h into the postoperative period. Nine RCTs with 632 patients were used for subanalysis of myocardial infarction, which was significantly lower in those who received epidural analgesia (rate difference = –3.8% with 95% CI of –7.4% to –0.2%; P = 0.049). This subgroup analysis on myocardial infarction was performed almost exclusively on vascular surgery patients (579 of 632 patients), and analgesic subgroup analysis revealed that TEA but not LEA provided a significant reduction in the rate of myocardial infarction (3.6% vs 8.5%, rate difference = –5.3% with 95% CI of –9.9% to –0.7%). A similar but more procedure-specific meta-analysis of open abdominal aortic surgery with 1224 patients (through June 2004) noted significant reduction in the risk of cardiovascular complications (RR 0.74 with 95% CI 0.56–0.97) and myocardial infarction (RR 0.52 with 95% CI 0.29–0.93) with epidural versus systemic analgesia (7). Subgroup analysis again indicated that only TEA and not LEA was associated with reduced risk of myocardial infarction. These findings would support the experimental data demonstrating physiologic cardiac benefits of TEA but not necessarily LEA. Another procedure-specific meta-analysis examined 15 RCTs (through January 2004) with 1178 patients undergoing CABG surgery with or without TEA (1). Although CABG surgery is typically considered high risk, incidences of myocardial infarction were less than the previous meta-analyses and reduction in incidence by TEA (2.3% vs 3.4%) did not reach statistical significance. The meta-analysis did note an inadequate sample size to examine differences in risk of myocardial infarction because of the low underlying incidence. A benefit of significant reduction in incidence of dysrhythmias was noted with TEA (17.8% vs 30%, OR 0.52 with 95% CI 0.29–0.93), again suggesting some benefit from a TEA technique. Other procedure-specific meta-analyses examining effects of epidural analgesia on abdominal and hip and knee replacement surgery concluded that there was insufficient evidence to analyze cardiovascular complications (10,11).
Randomized Controlled Trial.
Both the VACS and MASTER trials were previously described (Table 1). The VACS trial did not note a significant reduction in cardiovascular complications (myocardial infarction, heart failure, dysrhythmias, severe hypotension) with use of epidural morphine (8.6% vs 11.2%) for all patients (12). However, the abdominal aortic surgery subgroup (n = 374) had significantly lower incidences of cardiovascular complications (9.8% vs 17.9%, P = 0.03) primarily because of reduction in myocardial infarction (2.7% vs 7.9%, P = 0.05). This finding was somewhat unexpected, as patients received only epidural morphine and no local anesthetics for postoperative analgesia. However, there is some evidence that epidural analgesia (lumbar) per se may reduce cardiovascular complications (16) and that epidural morphine may attenuate postoperative sympathetic hyperexcitability (24). The MASTER trial also did not observe cardiovascular benefit from epidural analgesia (2.6% vs 2.4%) but was limited by poor protocol compliance in the epidural group, as only 225/447 patients fully completed the protocol. A subgroup analysis on aortic surgery patients (n = 164) was also performed but no significant differences in cardiovascular complications were noted (4.5% vs 4.7%) (25). The inconsistency between the MASTER and VACS subgroup analyses for aortic surgery may be explained by the much lower underlying incidence of cardiovascular complications and smaller number of aortic surgery patients in the MASTER trial. As such, the subgroup analysis is not adequately sized to detect a difference in cardiovascular outcomes.
Medicare Database.
As discussed earlier, a 5% random sample of 1997–2001 Medicare claims data for 12 mixed surgical procedures was examined in 2004 (Table 1). No differences in cardiovascular complications were observed between the groups with (n = 12,870) and without (n = 55,943) claims for epidural analgesia at 7 or 30 days. However, overall complication rates for myocardial infarction, angina, and heart failure were very low (0%–0.06%), and sample size was probably inadequate. Similar findings were reported in an earlier 5% random sample of only total hip replacement patients performed between 1994 and 1999 with no differences in cardiovascular complications between patients with (n = 2591) and without (n = 20,545) claims for epidural analgesia. Again, cardiovascular complication rates were quite low (0.8%–4%) (14).
Summary Statement.
There is minimal evidence that epidural analgesia reduces cardiovascular complications in the general surgical population. This is probably due to a low underlying rate of these complications in the general surgical population (0%–4%). There is consistent evidence that TEA may reduce the risk of cardiovascular complications, especially myocardial infarction, in patients undergoing major vascular surgery. This is likely due to a higher underlying rate of cardiovascular complications for this surgical population (4%–18%). Growth of minimally invasive vascular surgery procedures (e.g., endoluminal abdominal aortic aneurysm repair) with inherently lower cardiovascular complication rates may attenuate this finding (26).
Pulmonary.
Recent Guidelines (27) published by the American College of Physicians confirm that PPC remain a significant problem. PPCs are as common as cardiac complications for patients undergoing noncardiac procedures, and may carry the same risk of increased mortality and length of hospital stay (27). The pathophysiology of pulmonary dysfunction after surgery is multifactorial and may include disruption of normal respiratory muscle activity from either surgery or anesthesia, a reflex inhibition of phrenic nerve activity with subsequent decrease in diaphragmatic function, and uncontrolled postoperative pain which may contribute to postoperative loss of respiratory mechanics (28).
Epidural analgesia will confer superior analgesia compared with that from systemic opioids including IV PCA, which may improve voluntary pulmonary function (29). Segmental block from thoracic epidural anesthesia may result in increased tidal volume and vital capacity related in part to improved pain control and also to interruption of the reflex inhibition of phrenic nerve activity, thus improving diaphragmatic activity. However, effects from the typical dilute solutions of local anesthetics and opioid used for TEA are unclear. Indeed, the physiologic effects on respiratory muscle function are complex, and it has been suggested that TEA may potentially impair postoperative pulmonary function by paralyzing respiratory muscles such as the intercostals or abdominals (28). However, it has been demonstrated that TEA with bupivacaine 0.25% does not impair ventilatory mechanics, inspiratory respiratory muscle strength, or airway flow even in patients with severe chronic obstructive pulmonary disease (30,31), thus overall the effects of TEA would be expected to be beneficial.
Meta-Analysis.
Two meta-analyses were identified that examined the effects of epidural analgesia on PPC after mixed surgical procedures (Table 1). The CORTRA study in 2000 was the largest (n = 9559) and was described earlier. Use of neuraxial block (spinal anesthesia, epidural anesthesia, and epidural analgesia) in mixed surgical procedures was associated with a significantly decreased risk of pneumonia (3.1% vs 6%, OR 0.61 with 95% CI 0.48–0.76) especially with TEA (OR 0.48 with 95% CI 0.35–0.67) versus spinal anesthesia or lumbar epidural anesthesia (OR 0.76 with 95% CI 0.55–1.04) (6). This finding would support the underlying potential physiologic benefit of TEA for reducing PPCs. This finding is in agreement with an earlier meta-analysis (9) that more specifically examined 18 RCTs (through 1995 with 1016 patients for effects of epidural analgesia versus systemic analgesia (IV prn, IM prn, IV-PCA) on PPCs. This older meta-analysis noted a reduced risk of PPCs with epidural regimens using local anesthetics (RR 0.58 with 95% CI 0.42–0.8) and specifically for pulmonary infections (RR 0.36 with 95% CI 0.21–0.65).
More procedure-specific meta-analyses were also identified. Use of TEA in open abdominal aortic surgery (n = 861) was associated with a significantly decreased risk of respiratory failure (RR 0.63 with 95% CI 0.51–0.79) and a nonsignificant decrease for pneumonia (RR 0.64 with 95% CI 0.38–1.05) (7). Use of TEA in CABG surgery (n = 644) was associated with a significantly decreased risk of PPC (17.2% vs 30.3%, OR 0.41 with 95% CI 0.27–0.6) (1). Meta-analyses on RCTs examining the use of epidural analgesia in abdominal surgery and total hip and knee replacement surgery concluded that there were insufficient subjects to perform analysis on PPCs (10,11).
Randomized Controlled Trial.
Both the VACS and MASTER trials were described earlier (Table 1). The VACS study observed a nonsignificant reduction in respiratory failure for all patients in the epidural group (9.9% vs 14%) (12). However, subgroup analysis of the abdominal aortic surgery subgroup (n = 374) noted a significant reduction in respiratory failure with use of epidural analgesia (14% vs 28%, P < 0.01). The MASTER study (n = 915) found observed similar findings with a lower incidence of respiratory failure in the epidural analgesia group for high risk patients undergoing mixed abdominal surgical procedures (23% vs 30%, P = 0.02) (13). As described earlier, most epidurals were placed at the thoracic level for both RCTs.
Medicare Database.
As discussed earlier, a 5% random sample of 1997–2001 Medicare claims data for 12 mixed surgical procedures was examined in 2004. No differences in risk of pneumonia or respiratory failure were noted at 7 days (incidences 0%–0.15%), but an increased risk for pneumonia was associated with epidural anesthesia at 30 days (0.15% vs 0.09%, OR 1.91 with 95% CI 1.09–1.67). There was no obvious causative reason for this increased risk with epidural analgesia, and the finding may have been due to methodological limitations. An earlier 5% random sample of only total hip replacement performed between 1994 and 1999 reported no differences in incidences of pneumonia at 7 or 30 days between patients with (n = 2591) and without (n = 20,545) claims for epidural analgesia (2.3% vs 2.2%) (14).
Summary Statement.
There is consistent evidence from meta-analyses and large RCTs that TEA reduces risk of PPCs, especially in high-risk surgery such as abdominal aortic surgery and CABG. Again, growth of minimally invasive vascular surgery procedures may attenuate some of these findings (26).
Gastrointestinal.
Postoperative ileus is very common after abdominal surgery (>90% in many series) and may increase resource utilization (32). Several series in patients undergoing noncardiac surgery have observed that ileus was the most common issue in delaying hospital stay beyond 7 (51% of patients) and 10 days (42% of patients) (32). Although the pathophysiology of postoperative ileus and decreased gastrointestinal (GI) motility is multifactorial, primary mechanisms include neurogenic (spinal, supraspinal adrenergic pathways), inflammatory (i.e., local inflammatory responses initiate neurogenic inhibitory pathways), and pharmacologic (e.g., opioids) mechanisms (33).
Epidural analgesia with local anesthetics offers potential means to attenuate several mechanisms of postoperative ileus. Both postoperative pain and use of systemic opioids increase the risk of ileus. Epidural analgesia provides pain control superior to systemic opioids (including IV PCA) and allows marked sparing of opioid consumption (29). Sympathetic block from epidural local anesthetics may help attenuate postoperative reflex inhibition of GI motility. Suppression of the surgical stress response and systemic absorption of epidural local anesthetics may reduce the inflammatory response to attenuate postoperative ileus (32,33). Consistent with these mechanisms, experimental data consistently indicate that epidural analgesia with local anesthetics shortens time of intestinal paralysis, increases the strength of colonic contractions, and does not impair anastomotic healing or increase risk of anastomotic leakage (34).
Meta-Analysis.
A Cochrane Library meta-analysis was identified (updated through November 2003) that included 22 RCTs with 1023 patients undergoing abdominal surgery (Table 2) (35). In agreement with experimental data, this meta-analysis concluded that epidural analgesia with local anesthetics consistently showed reduced time to return of GI function compared with systemic opioids (–37 h with 95% CI –55 to –19 h) or epidural opioids (–24 h with 95% CI –38 to –10 h).
|
Randomized Controlled Trial.
No new RCTs since the above meta-analysis were identified that fit inclusion criteria.
Medicare Database.
As discussed earlier, a 5% random sample of 1997–2001 Medicare claims data for 12 mixed surgical procedures was examined in 2004. The incidence of ileus was extremely low (0% vs 0.007%) and no differences were observed between the groups with (n = 12,870) and without (n = 55,943) claims for epidural analgesia at 7 days (2). Similar findings were reported in an earlier 5% random sample of only total hip replacement performed between 1994 and 1999 with no differences in incidences (0.7% vs 1%) between patients with (n = 2591) and without (n = 20,545) claims for epidural analgesia (14).
Summary Statement.
There is consistent evidence from meta-analysis that epidural analgesia with local anesthetics hastens return of postoperative GI function after abdominal surgery by 24–37 h.
Coagulation.
Hypercoagulability occurs in association with surgical procedures with resultant risk of formation of deep venous thrombosis (DVT) and potentially fatal pulmonary embolism (PE). The field of surgical thromboprophylaxis is rapidly evolving, and multiple effective strategies have been identified. Current incidences of DVT and fatal PE in the presence of thromboprophylaxis are difficult to estimate because of use of different strategies. Older data in patients without thomboprophylaxis reported incidences for DVT of 0.4%–4% for general surgery, 1%–5% for urological surgery, 5%–36% for joint replacement, and 23%–30% for hip fracture surgery. Incidences for fatal PE were 0.1%–0.4% for general surgery, 0.5% for urological surgery, 0.1%–2% for joint replacement, and 2.5%–7.5% for hip fracture surgery (36). However, current surgical practice typically includes use of at least one form of thromboprophylaxis, and systematic reviews indicate that these interventions reduce risk of DVT and PE by 50%–80% (37). Thus, current incidences of postoperative DVT and PE are likely to be markedly less than those reported in older studies.
After surgery, the normal process of coagulation may become unbalanced, which may result in a tendency toward thrombosis. Immediately after surgical incision, there are increases in levels of tissue factor, tissue plasminogen activator, plasminogen activator inhibitor-1, and von Willebrand factor, which contribute to a hypercoagulable and hypofibrinolytic state postoperatively (38). Spinal and epidural anesthesia with local anesthetic regimens will attenuate perioperative hypercoagulability and may provide physiologic benefits (e.g., increased blood flow) to prevent perioperative coagulation-related complications. Intraoperative neuraxial anesthesia has been shown to attenuate perioperative increases in coagulation proteins and platelet activity, preserve fibrinolytic activity, and increase arterial and venous blood flow. For epidural anesthesia, there may be systemic absorption of local anesthetics which may exert systemic antithrombotic effects, including reduction in platelet aggregation, inhibition of thrombus formation, and reduction in blood viscosity (39). All these potential benefits are primarily gained from intraoperative neuraxial block, and whether extension of epidural analgesia into the postoperative period is important has been unclear. In fact, some experimental data suggest that postoperative epidural analgesia using common local analgesic concentrations (
0.125% bupivacaine) does not provide any significant increase in whole limb, venous, or cutaneous blood flow nor a decrease in postoperative hypercoagulability (40).
Meta-Analysis.
The previously described CORTRA meta-analysis (RCTs up to January 1, 1997) noted significant reduction in risk of DVT (2.9% vs 4.7%) and PE (0.6% vs 1.4%) with use of neuraxial block (Table 2) (6). However, the effects of epidural analgesia per se are unclear as all neuraxial block patients received intraoperative regional anesthesia, and type of regional block was mixed between spinal and epidural blocks. In addition, the meta-analysis did not comment on use of thromboprophylaxis, and many of the underlying RCTs were performed before the release of currently popular thromboprophylactic drugs (only 38/141 RCTs published >1990). Procedure-specific meta-analyses for open aortic surgery, abdominal surgery, and total hip and knee replacement concluded there were insufficient subjects for analysis (7,10,11).
Randomized Controlled Trial.
No additional RCTs were identified that met inclusion criteria.
Medicare Database.
As discussed earlier, a 5% random sample of 1997–2001 Medicare claims data for 12 mixed surgical procedures was examined in 2004 (Table 2). Incidences of DVT (0.02% vs 0.01%) and PE (0.04% vs 0.03%) were extremely low and no differences were observed between the groups with (n = 12,870) and without (n = 55,943) claims for epidural analgesia at 7 days (2). An earlier 5% random sample of only total hip replacement performed between 1994 and 1999 reported an increased incidence of DVT (0.97% vs 0.61%, P = 0.04) but similar incidence of PE (0.5% vs 0.6%) at 7 days in patients with (0.7%, n = 2591) and without (1%, n = 20,545) claims for epidural analgesia (14). The increased incidence of DVT in the epidural group is puzzling and was explained by the authors as likely due to methodological flaws of the database (e.g., poor coding of complications).
Summary Statement.
There is minimal evidence that postoperative epidural analgesia affects risk of DVT and PE, and very few studies have addressed this question with use of current methods of effective thromboprophylaxis.
Other Outcomes
Infectious Complications.
After major surgical procedures, there is an early hyperinflammatory response with release of proinflammatory tumor necrosis factor 
, interleukin-1 and interleukin-6 cytokine, neutrophil activation and microvascular adherence, and uncontrolled polymorphonuclear and macrophage oxidative activity. This ultimately results in significant cell-mediated immunosuppression marked by monocyte deactivation, decreased microbicidal activity of phagocytes, and an overall imbalance between proinflammatory and antiinflammatory cytokines and immunocompetent cells (41). Epidural analgesia may reduce infectious complications by reducing lymphocyte suppression, attenuating proinflammatory cytokines, and by increasing surgical wound oxygen tension to promote healing (42,43), although such positive effects may be negated by the more extensive inflammatory response from more extensive surgery (44).
Evidence:
The previously described CORTRA meta-analysis noted few incidences of wound infections (0.05% vs 0.07%) without differences between groups. No other meta-analyses or RCTs were identified that met inclusion criteria.
Cognitive Decline and Delirium.
Postoperative cognitive decline is common. Mental function typically reaches a nadir in the early postoperative period, with a recovery to preoperative levels by 1 wk after surgery in most patients. The elderly are at the highest risk of postoperative cognitive decline (7%–26%) and delirium (10%–60%) after noncardiac surgery (45). Delirium is a particularly problematic subset of cognitive dysfunction, as it may contribute to an additional 17.5 million inpatient days and 4 billion USD in health care expenditures annually in hospitalized patients (46). The etiology of postoperative cognitive decline and delirium is not well understood and is probably multifactorial. The severity of postoperative pain and use of IV opioids have been identified as risk factors for postoperative delirium in the elderly (45,47). As epidural analgesia provides better analgesia and spares systemic opioids (29), there is a theoretical reason for improved cognitive outcomes with epidural analgesia.
Evidence:
No meta-analyses or RCTs were identified that met inclusion criteria. A qualitative systematic review was published in 2006 (45) that identified four small RCTs (n = 30–70) and one case–control study that evaluated cognitive decline and delirium in patients divided between epidural versus systemic analgesia. Although no differences between groups were noted, the review commented on the small sample sizes and poor study quality of the individual studies.
Chronic Pain.
Chronic pain after surgery has only recently been considered as a significant postoperative complication. The incidence of chronic pain after surgery varies by procedure, but may be quite high, with the incidence ranging from 30%–81% after limb amputation, 22%–67% after thoracotomy, 17%–57% after breast surgery, and 4%–37% after hernia repair (48,49). The etiology of chronic postoperative pain is unclear but may include peripheral and central sensitization among other factors. Sensitization may explain why the severity of acute postoperative pain is a common risk factor for development of chronic postoperative pain (48). Theoretically, use of epidural analgesia would confer superior postoperative analgesia and possibly preemptive analgesia (29,50), which might result in a lower incidence of chronic postsurgical pain.
Evidence:
No meta-analyses or RCTs were identified that met inclusion criteria. One meta-analysis (51) published in 2005 examined the effects of preemptive versus postoperative epidural analgesia for postthoracotomy pain. One qualitative systemic review published in 2004 (52) identified one RCT (n = 69) that examined epidural versus systemic analgesia for postthoracotomy pain and three RCTs that examined preemptive epidural analgesia for postamputation pain (n = 30–60).
Summary Statement.
There is minimal evidence that epidural analgesia affects risk of wound infection, postoperative cognitive decline and delirium, or development of chronic postoperative pain, as very few studies have specifically addressed these questions.
Continuous Peripheral Nerve Analgesia
Continuous peripheral techniques offer potential advantages over systemic analgesic regimens for better analgesia with fewer side effects than opioid-based regimens (53). Peripheral techniques offer more targeted and limited sensory and motor block than central neuraxial analgesia and have less risk of catastrophic complications such as spinal hematomas and abscesses. However, the attractive limitation of conduction block with perineural analgesia limits the potential effects on mechanisms of postoperative complications. For example, peripheral limitation of block typically minimizes the effects on the sympathetic nervous system (54), which is thought to be an important contributor to cardiovascular (20) and GI (32) morbidity, and limits effects on the phrenic nerve, which may be a contributor to improved pulmonary function (28). Another limitation of ability of the peripheral nerve blocks to affect outcome is the frequent use of adjunct analgesics such as nonsteroidal antiinflammatory drugs (NSAIDs) or opioids. This multimodal approach may benefit the patient, but it lessens the potential impact of perineural analgesia on outcome per se. Finally, continuous perineural analgesia is typically applied for low-risk surgery, such as elective orthopedic or ambulatory procedures, which have intrinsically low rates of postoperative complications which are unlikely to be altered (55).
Mortality and Major Morbidity.
Evidence:
A meta-analysis (53) was published in 2006 of all available RCTs (through May 2004) comparing continuous perineural analgesia to systemic opioids included 19 RCTs with a total of 603 patients. Analgesia and side effects were the primary outcomes. Included RCTs were small (n = 20–62), and no comments were made on mortality or major morbidity. No other RCTs were identified that met inclusion criteria.
Summary Statement.
There is minor theoretical and minimal clinical evidence that perineural analgesia affects postoperative mortality or morbidity, as essentially no RCTs have addressed this question.
Continuous Wound Catheters
A promising modality is directly placing catheters to infuse local anesthetics into wounds. This modality improves analgesia compared with placebo, and reduces the need for opioids and thus their drug-related side effects (56). In addition to analgesia, continuous application of local anesthetic to wounds may affect postoperative outcome via other mechanisms. For example, studies have observed that direct applications of local anesthetics reduce release of inflammatory mediators from neutrophils, reduce neutrophil adhesion to the endothelium, reduce formation of free oxygen radicals, and decrease edema formation (57,58). In addition, previous studies have reported appreciable, though nontoxic, blood levels of local anesthetic from the use of continuous wound catheters (56). Systemic administration of local anesthetics has been shown to provide postoperative analgesia, reduce opioid consumption, and reduce ileus after abdominal surgery (59,60).
Mortality and Major Morbidity.
Evidence:
A meta-analysis was published in 2006 that included 43 RCTs (through February 2006) with 2058 patients (56). Analgesia and side effects were the primary outcomes. Included RCTs were small (n = 20–86), and no comments were made on mortality or major morbidity. No other RCTs were identified that met inclusion criteria.
Summary Statement.
There is minor theoretical and minimal clinical evidence that continuous wound catheter analgesia affects postoperative mortality or morbidity, as essentially no RCTs have addressed this question.
Systemic Analgesia
IV PCA with Opioids
Because of its ability to titrate to individual needs, IV PCA is considered by many as the "gold standard" for delivery of IV opioid for the management of postoperative pain (29). However, IV PCA offers limited ability to affect postoperative outcomes when compared with conventional systemic delivery. The most recent systematic reviews of RCTs (through January 2000 and November 2004) did not note substantial differences in analgesia, side effects, or opioid use with IV PCA compared with conventional delivery of opioids (61,62). The primary benefit was some modest and inconsistent increase inpatient satisfaction with use of IV PCA. Overall, use of IV PCA results in a very similar analgesic profile to conventional systemic delivery and offers little theoretical avenues to affect postoperative outcome.
Mortality and Major Morbidity.
Evidence:
The two most recent meta-analyses (61,62) examining the efficacy of IV PCA opioids (versus conventional PRN opioid administration) on perioperative outcomes in patients undergoing mixed surgery were published in 2005 and 2001 and included a total of 56 (up to November 2004) and 22 (up to January 2000) RCTs with 3861 and 1139 patients (Table 3). Not surprisingly, the primary outcomes evaluated in the RCTs were analgesic efficacy, analgesic usage, opioid-related side effects, and hospital stay. There was in general, insufficient data to allow quantitative analysis for mortality and major morbidity including cardiovascular, GI, or coagulation-related outcomes. However, the 2001 meta-analysis did note that PPCs from two RCTs were less with IV PCA (70/75, or 93.3%) versus control (77/77 or 100%) with a RR = 1.07 (95%CI 1.01–1.14). These are very high incidences of PPC regardless of technique and do not suggest a clinically relevant decrease in risk with use of IV PCA. No additional RCTs were identified for inclusion.
|
|
Multimodal Analgesics
NSAIDs, cyclooxygenase 2 inhibitors (COX2I), acetaminophen, and ketamine are considered multimodal analgesics that all act through nonopioid analgesic pathways. Four meta-analyses (63,64,66,67) were most recently published in 2005 and 2006 with 52 (through July 2004), 22 (through December 2003), 7 (through April 2003), and 37 (through June 2004 and unpublished data) RCTS with 4893, 2307, 591, and 2240 patients undergoing mixed surgical procedures have evaluated analgesic efficacy of these drugs, and all consistently identified significant opioid sparing with all drugs. However, only NSAIDs and ketamine were noted to moderately reduce pain scores and opioid-related side effects. An additional qualitative and quantitative systematic review (69) was also published in 2005 that included 22 RCTs (through August 2004) with 2246 patients that evaluated the effects of preoperative administration of COX2I. This systematic review also found opioid sparing and some reduction in postoperative pain scores without a consistent decrease in side effects. Overall, these drugs do not offer dramatic changes in analgesic profiles when added to systemic opioids, and thus offer minimal theoretical mechanisms for improved postoperative outcome. Both NSAIDs and COX2I may provide some additional theoretical mechanisms due to the antiinflammatory activities.
Mortality and Major Morbidity.
Evidence:
None of the above meta-analyses assessed postoperative mortality or previously discussed morbidity as primary outcomes. The largest meta-analysis (63) noted increased risk of severe bleeding with NSAIDs (0% vs 1.7%, number needed to harm = 59) and increased risk of renal failure with COX2I (0% vs 1.4%, number needed to harm = 73). Although not quantitatively addressed in these meta-analyses, COX2I drugs have been significantly associated with increased risk of cardiovascular complications in the nonsurgical population, and rofecoxib has been withdrawn (70). One meta-analysis (65) published in 2005 that included only 3 RCTs (through February 2005) with 2604 subjects undergoing primarily cardiac surgery also noted an increased risk of cardiovascular complications with valdecoxib/paracoxib versus placebo (2.6% vs 0.9%, OR 2.3 with 95% CI 1.1–4.7). An additional RCT published in 2006 (68) randomized 1062 patients to COX2I (valdecoxib/paracoxib) or placebo in addition to systemic opioid after mixed noncardiac surgical procedures. Use of COX2I decreased opioid consumption and pain scores but did not affect risk of mortality, cardiovascular complications, renal failure, or wound infection. All complications of interest were infrequent (<1%), and thus sample size was insufficient for this purpose. No additional RCTs were identified that met inclusion criteria. One additional RCT published in 2005 (71) randomized 1003 patients undergoing mixed surgical procedures to a single dose of either ketorolac or morphine and thus is not included.
Summary Statement.
There is minimal theoretical and insufficient quantitative data to determine if addition of multimodal analgesics to systemic opioids results in improvement in perioperative mortality or major morbidity. There is some evidence from meta-analyses that use of NSAIDs and COX2I may increase risk of severe bleeding, renal failure, and cardiovascular complications.
CONCLUSION AND FUTURE DIRECTIONS
Overall, there is insufficient evidence to confirm or deny the ability of postoperative analgesic techniques to affect major postoperative mortality or morbidity. Epidural analgesia with local anesthetics has the greatest theoretical potential to affect outcome and correspondingly has been the most thoroughly investigated technique. The majority of evidence favors an ability of epidural analgesia to reduce postoperative cardiovascular complications and PPC after major vascular surgery and in high-risk patients. This finding may become irrelevant due to rapid conversion of major surgery to minimally invasive techniques (e.g., endoluminal abdominal aortic repair) that have less risk of complications (26). There is also consistent evidence that epidural analgesia with local anesthetics is associated with faster resolution of postoperative ileus after major abdominal surgery. Again, this finding may also become irrelevant with adoption of laparoscopic techniques and multimodal fast-track programs for abdominal surgery (72). There is no current evidence that perineural analgesia, continuous wound catheters using local anesthetics, IV PCA with opioids, or addition of multimodal systemic analgesics have any clinically significant beneficial effect on postoperative complications.
A primary finding of this review is the small number of patients who have been enrolled for study. This is a critical limitation of nearly all discussed studies due to the relatively low underlying incidences of complications. Incidences of mortality ranged from 0.2% to 5% and of serious cardiopulmonary complications from 0%–18%, depending on surgical procedure and patient population. Such incidences require subject samples ranging from 500 to 50,000 to detect even a 50% reduction in incidence. Not surprisingly, the higher risk procedures have more frequent complications, and it is these populations in which differences have been noted with analgesic techniques. However, risk of complications is also rapidly declining in these very same surgical procedures because of changes in medical practice. As surgery becomes increasingly safer with adoption of minimally invasive procedures and with better perioperative care, it will become more difficult to assess the effects of individual interventions on outcomes because of the underlying low incidences. Thus, the future will require increasingly large RCTs with sufficient patient numbers.
It may be difficult to determine an exact number of subjects required for these RCTs, as some of the variables used for sample size calculations (e.g., hypothesized difference between groups or standard deviation) are in themselves estimates of relatively uncommon events. The accuracy of these estimates may affect the validity of the sample size calculations to the point that calculated sample sizes may still be inadequate. Previous studies have noted that up to 25% of RCTs that reported sample size calculations were still underpowered because of poorly estimated parameters used for the calculation (73). This potential difficulty in predicting adequate sample size is an underlying reason for the common recommendation to provide confidence intervals when reporting primary and secondary outcomes in RCTs (www.consort-statement.org). Confidence intervals are especially valuable for statistically insignificant differences, as they often indicate that the result does not exclude an important clinical difference. In terms of other study designs, it appears clear that individual small RCTs will not answer questions concerning mortality and major morbidity by themselves, although they may add to a future meta-analysis. Alternatively, design and implementation of large observational databases may also allow the acquisition of large amounts of patient data, but with accompanying methodological limitations. Such an endeavor is currently underway at the American Society of Regional Anesthesia with the Acute Postoperative Pain study group (74).
With the increasing safety of surgical procedures, it may be time to increase the scope of definition of "outcomes" of interest. As mortality and major morbidity become increasingly uncommon, perhaps it will be appropriate to focus more upon patient-oriented outcomes. These nontraditional outcomes are assessed from the patient’s perspective, which mirrors the increasing importance of patient-oriented assessments in other areas of medicine and increasing consumerism in health care (75). Domains typically comprise quality of postoperative recovery, including analgesia and side effects, patient satisfaction, and quality of life (76). Interestingly, patients are often more concerned about these common, low morbidity outcomes, whereas anesthesiologists are often more concerned about major complications that are increasingly uncommon (75).
Finally, it has become recognized that analgesia is probably a single component in a comprehensive postoperative rehabilitation program. Many other factors (e.g., early mobilization, early oral nutrition, fluid balance) have also been shown to be important therapeutic components and will need to be integrated into future studies examining major or patient-oriented outcomes (77).
Footnotes
Dr. Liu, Section Editor for Pain Medicine, was recused from all editorial decisions related to this manuscript.
Accepted for publication November 15, 2006.
Reprints will not be available from the author.
REFERENCES
